Probabilistic topic modelling in food spoilage analysis: A case study with Atlantic salmon (Salmo salar).

Probabilistic topic modelling is frequently used in machine learning and statistical analysis for extracting latent information from complex datasets. Despite being closely associated with natural language processing and text mining, these methods possess several properties that make them particularly attractive in metabolomics applications where the applicability of traditional multivariate statistics tends to be limited. The aim of the study was thus to introduce probabilistic topic modelling - more specifically, Latent Dirichlet Allocation (LDA) - in a novel experimental context: volatilome-based (sea) food spoilage characterization. This was realized as a case study, focusing on modelling the spoilage of Atlantic salmon (Salmo salar) at 4 °C under different gaseous atmospheres (% CO2/O2/N2): 0/0/100 (A), air (B), 60/0/40 (C) or 60/40/0 (D). First, an exploratory analysis was performed to optimize the model tunings and to consequently model salmon spoilage under 100% N2 (A). Based on the obtained results, a systematic spoilage characterization protocol was established and used for identifying potential volatile spoilage indicators under all tested storage conditions. In conclusion, LDA could be used for extracting sets of underlying VOC profiles and identifying those signifying salmon spoilage, giving rise to an extensive discussion regarding the key points associated with model tuning and/or spoilage analysis. The identified compounds were well in accordance with a previously established approach based on partial least squares regression analysis (PLS). Overall, the outcomes of the study not only reflect the promising potential of LDA in spoilage characterization, but also provide several new insights into the development of data-driven methods for food quality analysis.

Spoilage evaluation of raw Atlantic salmon (Salmo salar) stored under modified atmospheres by multivariate statistics and augmented ordinal regression.

The development of quality monitoring systems for perishable food products like seafood requires extensive data collection under specified packaging and storage conditions, followed by advanced data analysis and interpretation. Even though the benefits of using volatile organic compounds as food quality indices have been recognized, few studies have focused on real-time quantification of the seafood volatilome and subsequent systematic identification of the most important spoilage indicators. In this study, spoilage of Atlantic salmon (Salmo salar) stored under modified atmospheres (% CO2/O2/N2) and air was characterized by performing multivariate statistical analysis and augmented ordinal regression modelling for data collected by microbiological, chemical and sensory analyses. Out of 25 compounds quantified by selected-ion flow-tube mass spectrometry, ethanol, dimethyl sulfide and hydrogen sulfide were found characteristic under anaerobic conditions (0/0/100 and 60/0/40), whereas spoilage under air was primarily associated with the production of alcohols and ketones. Under high-O2 MAP (60/40/0), only 3-methylbutanal fulfilled the identification criteria. Overall, this manuscript presents a systematic and widely applicable methodology for the identification of most potential seafood spoilage indicators within the context of intelligent packaging technology development. In particular, parallel application of statistics and modelling was found highly beneficial for the performance of the quality characterization process and for the practical applicability of the obtained results in food quality monitoring.

Tracking the sources of psychrotrophic bacteria contaminating chicken cuts during processing.

The major aim of the study was to establish the routes via which spoilage associated psychrotrophic bacteria contaminate poultry products at a large processing plant located in Belgium. Environmental samples were collected consisting of samples of air and swabs of food contact surfaces. Product samples were also collected consisting of modified atmosphere packaged (MAP) chicken wings and legs, which were analyzed microbiologically on the same day they were produced as well as after their sell-by date. Psychrotrophic bacteria from these samples were subsequently clustered and identified by means of MALDI-TOF MS and 16S rRNA gene sequencing. Carnobacterium maltaromaticum was determined to dominate the spoilage flora of both wings and legs. Other psychrotrophic bacteria able to grow on MRS which were identified on expired wings and legs included Carnobacterium divergens, Brocothrix thermosphacta, Lactobacillus curvatus, and Lactobacillus brevis. These were determined to arise from food contact surfaces such as cutting blades, leg hooks, Ertalon and polyurethane conveyor belts, working tables, and the hands of the operators. Importantly, it was determined that cleaning and disinfection was largely inadequate. Air was also determined to be an important vector of psychrotrophic bacteria in the processing environment, potentially contaminating the products directly or indirectly.

Solid fat influences sorbic acid partitioning and enhances the preservation effect on C. guilliermondii in biphasic food model systems.

Preservation of emulsions relies on factors including pH, temperature, structure and the application of carboxylic acid preservatives, such as sorbic and benzoic acid. Organic acid preservatives tend to migrate to the lipid phase of emulsions. Taking into account the fact that organic acid in the aqueous phase is solely responsible for antimicrobial activity, this partitioning behavior is considered as a loss of preservative. The influence of microstructure properties on sorbic acid distribution and preservation effect was investigated in model food systems comprising of aqueous phase, and liquid oil and solid fat as lipid phase, which represent major constituents of water-oil emulsions. The aqueous phase of the food model systems was comprised of Yeast Nitrogen Broth (YNB) in phosphate buffer and buffered at pH 4.5, 5.5, 6.5. Sorbic acid (100 mg/L) in the form of potassium sorbate was added to the aqueous phase. Candida guilliermondii (2 log CFU/mL) was inoculated in the aqueous phase to resemble industrial post-contamination CFU levels. Growth parameters, generation time (GT) and lag phase (λ) of C. guilliermondii in the aqueous phase of the food model systems were quantified during 1 month at 7 °C and maximum 80 h at 22 °C. HPLC analyses were performed to evaluate total sorbic acid content in each food model system. Sorbic acid inhibited growth of C. guilliermondii in YNB + SA at pH 4.5 at 7 °C and 22 °C. The presence of liquid oil caused partitioning of sorbic acid into the lipid phase in YNB + SA|oil at pH 4.5 at 7 °C and 22 °C, reducing its inhibitory effect. Adding solid fat into the model food systems significantly prolonged lag phase duration in the YNB + SA|oil + fat system at pH 4.5 at 22 °C and growth was inhibited at 7 °C. HPLC analysis showed a final aqueous sorbic acid concentration of 34 mg/L and 44 mg/L in YNB + SA|oil system at pH 4.5 at 7 °C and 22 °C after 1 month and 168 h, respectively. In YNB + SA|oil + fat system at pH 4.5, 87 mg/L of aqueous sorbic acid was measured after 1 month at 7 °C and 168 h at 22 °C, indicating that the presence of solid fat retards sorbic acid partitioning. Results show that structural components such as liquid oil and solid fat have an effect on the aqueous sorbic acid concentration and its preservation effect. The presence of solid fat reduces the tendency of sorbic acid to partition into the lipid phase, which is reflected in the inhibitory effect of sorbic acid on C. guilliermondii.

Microbiological, chemical and sensory spoilage analysis of raw Atlantic cod (Gadus morhua) stored under modified atmospheres.

During fish spoilage, microbial metabolism leads to the production of volatile organic compounds (VOCs), characteristic off-odors and eventual consumer rejection. The aim of the present study was to contribute to the development of intelligent packaging technologies by identifying and quantifying VOCs that indicate spoilage of raw Atlantic cod (Gadus morhua) under atmospheres (%v/v CO2/O2/N2) 60/40/0, 60/5/35 and air. Spoilage was examined by microbiological, chemical and sensory analyses over storage time at 4 or 8 °C. Selected-ion flow-tube mass spectrometry (SIFT-MS) was used for quantifying selected VOCs and amplicon sequencing of the 16S rRNA gene was used for the characterization of the cod microbiota. OTUs classified within the Photobacterium genus increased in relative abundance over time under all storage conditions, suggesting that Photobacterium contributed to spoilage and VOC production. The onset of exponential VOC concentration increase and sensory rejection occurred at high total plate counts (7-7.5 log). Monitoring of early spoilage thus calls for sensitivity for low VOC concentrations.

Antifungal properties of fermentates and their potential to replace sorbate and propionate in pound cake.

The major objective of this study was to assess the antifungal activities of commercially available 'clean label' fermentates and their potential to replace the preservative function of sorbate and propionate in cake. This study was performed in two parts. In the first part of the study the inhibitory activities of selected fermentates - FA, FB, FC and FD - towards Aspergillus tritici and Aspergillus amstelodami were assessed as a function of pH (5.0-6.5) on malt extract agar (MEA). In the second part of the study, challenge, shelf-life and sensorial tests were used to determine the suitability of these fermentates to replace potassium sorbate and calcium propionate in quarter pound cake. All the fermentates evaluated in this study all had significant (p<0.05) inhibitory activities towards A. tritici and A. amstelodami within the recommended dosage range for application in bakery products. In all cases, the inhibitory activity of the fermentates increased with a decrease in the pH and an increase in concentration. FC was generally the most inhibitory whilst FD was the least inhibitory. Significant (p<0.05) synergistic interactions were determined to occur between the effects of pH and concentration for all fermentates evaluated in this study. The sensorial tests with FC showed that cakes produced with ≤1% FC (on basis of the batter) did not differ significantly (p>0.05) in flavour from the reference cake (0.5% calcium propionate and 0.54% potassium sorbate). However, the challenge and shelf-life tests showed that cakes produced with ≤1% FC were not as microbiologically shelf-stable as the reference cake, especially when sliced. Therefore, it can be concluded that whilst fermentates have appreciable antifungal effects, their use could potentially result in reduced shelf-stabilities. Robust challenge and shelf-life tests would be recommended before the marketing of cakes were propionate and/or sorbate has been replaced to ensure accurate shelf-lives are stated.

Isoepoxydon dehydrogenase (idh) gene expression in relation to patulin production by Penicillium expansum under different temperature and atmosphere.

Penicillium expansum growth and patulin production occur mainly at post-harvest stage during the long-term storage of apples. Low temperature in combination with reduced oxygen concentrations is commonly applied as a control strategy to extend apple shelf life and supply the market throughout the year. Our in vitro study investigated the effect of temperature and atmosphere on expression of the idh gene in relation to the patulin production by P. expansum. The idh gene encodes the isoepoxydon dehydrogenase enzyme, a key enzyme in the patulin biosynthesis pathway. First, a reverse transcription real-time PCR (RT-qPCR) method was optimized to measure accurately the P. expansum idh mRNA levels relative to the mRNA levels of three reference genes (18S, ß-tubulin, calmodulin), taking into account important parameters such as PCR inhibition and multiple reference gene stability. Subsequently, two P. expansum field isolates and one reference strain were grown on apple puree agar medium (APAM) under three conditions of temperature and atmosphere: 20 °C - air, 4 °C - air and 4 °C - controlled atmosphere (CA; 3% O2). When P. expansum strains reached a 0.5 and 2.0 cm colony diameter, idh expression and patulin concentrations were determined by means of the developed RT-qPCR and an HPLC-UV method, respectively. The in vitro study showed a clear reduction in patulin production and down-regulation of the idh gene expression when P. expansum was grown under 4 °C - CA. The results suggest that stress (low temperature and oxygen level) caused a delay of the fungal metabolism rather than a complete inhibition of toxin biosynthesis. A good correlation was found between the idh expression and patulin production, corroborating that temperature and atmosphere affected patulin production by acting at the transcriptional level of the idh gene. Finally, a reliable RT-qPCR can be considered as an alternative tool to investigate the effect of control strategies on the toxin formation in food.

Methodology for modeling the disinfection efficiency of fresh-cut leafy vegetables wash water applied on peracetic acid combined with lactic acid.

A methodology to i) assess the feasibility of water disinfection in fresh-cut leafy greens wash water and ii) to compare the disinfectant efficiency of water disinfectants was defined and applied for a combination of peracetic acid (PAA) and lactic acid (LA) and comparison with free chlorine was made. Standardized process water, a watery suspension of iceberg lettuce, was used for the experiments. First, the combination of PAA+LA was evaluated for water recycling. In this case disinfectant was added to standardized process water inoculated with Escherichia coli (E. coli) O157 (6logCFU/mL). Regression models were constructed based on the batch inactivation data and validated in industrial process water obtained from fresh-cut leafy green processing plants. The UV254(F) was the best indicator for PAA decay and as such for the E. coli O157 inactivation with PAA+LA. The disinfection efficiency of PAA+LA increased with decreasing pH. Furthermore, PAA+LA efficacy was assessed as a process water disinfectant to be used within the washing tank, using a dynamic washing process with continuous influx of E. coli O157 and organic matter in the washing tank. The process water contamination in the dynamic process was adequately estimated by the developed model that assumed that knowledge of the disinfectant residual was sufficient to estimate the microbial contamination, regardless the physicochemical load. Based on the obtained results, PAA+LA seems to be better suited than chlorine for disinfecting process wash water with a high organic load but a higher disinfectant residual is necessary due to the slower E. coli O157 inactivation kinetics when compared to chlorine.

A new protocol for evaluating the efficacy of some dispensing systems of a packaging in the microbial protection of water-based preservative-free cosmetic products.

OBJECTIVE: A new protocol is described for assessing the efficacy of the dispenser of some packaging systems (PSs) of preservative-free cosmetic products in protecting both their contained formula and their delivered doses. METHODS: Practically, aiming at mimicking contacts with a non-sterile skin or fingers, the dispensing system is put into contact with a pre-contaminated fabric by a standardized colonization of P. aeruginosa. RESULTS: When applied to three different types of packaging, results show clear differences in both criteria between these conditioning articles, that is variable efficacies in protecting the contained product and the delivered doses, knowing that the first aspect is of paramount importance. CONCLUSION: The proposed protocol is proved being able to discriminate between different PSs and provides information on strong and weak features of certain types dispensing technologies prone to efficiently decrease either the dose contamination or to prevent contamination in reaching the contained product. Therefore, the proposed protocol can contribute to an objective selection of a PS for protecting a cosmetic care product with a low content of preservative or preservative free.

Modelling of the growth/no growth interface of Wallemia sebi and Eurotium herbariorum as a function of pH, aw and ethanol concentration.

High sugar products (sugar content > 50%) are generally considered to be stable against all forms of microbial spoilage during a prolonged shelf life of several months. However, one specific subgroup of micro-organisms, the xerophilic moulds, can develop quite fast on the surface of food products with a reduced water activity (< 0.85). The chance whether these xerophilic moulds are able to grow on the food product depends on the combination of intrinsic factors (e.g., water activity and pH) and the storage conditions (e.g., temperature). This study examines the development of growth/no growth models for the xerophilic moulds Wallemia sebi and Eurotium herbariorum in a sugar rich broth. Growth/no growth models are predictive models that are designed to give a prediction about the probability of growth of a spoilage micro-organism under a specific set of environmental conditions. In this research, a water activity between 0.75 and 0.90, a pH between 5.0 and 6.2, an ethanol concentration between 0% and 5% (g EtOH/g H2O) and their interactions were tested. The inoculated media were stored at 22 °C (± 1 °C) during a prolonged test period (up to 120 days). The obtained models were also validated in a chocolate-based food product (ganache). The resulting growth/no growth models show that the growth of W. sebi and E. herbariorum can be inhibited for a prolonged time (> 3 months) if an ethanol concentration of 5% on the water phase is present in the food product, irrespective of water activity values between 0.89 and 0.755. The necessary amount of ethanol for shorter shelf lives can be calculated with the models that were built. Although the models have not been validated thoroughly in actual food products, the preliminary results that were obtained by testing the model on a ganache indicated that the models are capable of delivering safe predictions.

Strategies to increase the stability of intermediate moisture foods towards Zygosaccharomyces rouxii: the effect of temperature, ethanol, pH and water activity, with or without the influence of organic acids.

Intermediate moisture foods (IMF) are in general microbiologically stable products. However, due to health concerns consumer demands are increasingly forcing producers to lower the fat, sugar and preservatives content, which impede the stability of the IMF products. One of the strategies to counteract these problems is the storage of IMF products at lower temperatures. Thorough knowledge on growth/no growth boundaries of Zygosaccharomyces rouxii in IMF products, also at different storage temperatures is an important tool for ensuring microbiologically stability. In this study, growth/no growth models for Z. rouxii, developed by Vermeulen et al. (2012) were further extended by incorporating the factor temperature. Three different data sets were build: (i) without organic acids, (ii) with acetic acid (10,000 ppm on product basis) and (iii) with sorbic acid (1500 ppm on product basis). For each of these data sets three different growth/no growth models were developed after 30, 60 and 90 days. The results show that the influence of temperature is only significant in the lower temperature range (8-15 °C). Also, the effect of pH is negligible (pH 5.0-6.2) unless organic acids are present. More specific, acetic acid had only an additive effect to ethanol and aw at low pH, whereas sorbic acid had also an additive effect at the higher pH values. For incubation periods longer than 30 days the growth/no growth boundary remained stable but enlarged gradually between day 60 and 90, except for the lower temperature range (<12 °C) where the boundary shifts to more stringent environmental conditions.

Incidence, diversity and characteristics of spores of psychrotolerant spore formers in various REPFEDS produced in Belgium.

The major objectives of this study were to determine the incidence of psychrotolerant spore formers from REPFEDS marketed in Belgium, and their diversity and characteristics. Spore formers in general were found as spores on 38.3% of the food samples and in 85% food products types evaluated. 76% of the food samples containing spore formers had spores before enrichment. A total of 86 spore formers were isolated from the samples. 28 of 86 bacterial spore formers (32.6%) were capable of vegetative growth at 7 °C. 96% (27/28) of these psychrotolerant spore formers were determined to belong to Bacillus or related genera. According to a (GTG)5-PCR analysis, 24 of these 28 isolates were genetically distinct from each other. 10.7% (3/28) of the bacilli were determined to belong to the Bacillus cereus group, namely B. cereus (chicken curry and Edam cheese) and Bacillus mycoides (Emmental cheese). Almost half of the bacilli (12/27) were putatively identified as Bacillus pumilus, which occurs ubiquitously in nature and has been associated with outbreaks of foodborne disease. Only one psychrotolerant clostridium, Clostridium tyrobutyricum, was isolated in the study. The results of this study show the highly diverse ecology and spoilage potential of psychrotolerant spore formers in REPFEDs marketed in Belgium.

The combined effect of pasteurization intensity, water activity, pH and incubation temperature on the survival and outgrowth of spores of Bacillus cereus and Bacillus pumilus in artificial media and food products.

The objective of the study was to evaluate the combined effects of pasteurization intensity (no heat treatment and 10 min at 70, 80 and 90 °C), water activity (aw) (0.960-0.990), pH (5.5-7.0) and storage temperature (7 and 10 °C) on the survival and outgrowth of psychrotolerant spores of Bacillus cereus FF119b and Bacillus pumilus FF128a. The experiments were performed in both artificial media and a validation was performed on real food products (cream, béchamel sauce and mixed vegetable soup). It was determined that in general, heat treatments of 10 min at 70 °C or 80 °C activated the spores of both B. cereus FF119b and B. pumilus FF128a, resulting in faster outgrowth compared to native (non-heat treated) spores. A pasteurization treatment of 10 min at 90 °C generally resulted in the longest lag periods before outgrowth of both isolates. Some of the spores were inactivated by this heat treatment, with more inactivation being observed the lower the pH value of the heating medium. Despite this, it was also observed that under some conditions the remaining (surviving) spores were actually activated as their outgrowth took place after a shorter period of time compared to native non-heated spores. While the response of B. cereus FF119b to the pasteurization intensity in cream and béchamel sauce was similar to the trends observed in the artificial media at 10 °C, in difference, outgrowth was only observed at 7 °C in both products when the spores had been heated for 10 min at 80 °C. Moreover, no inactivation was observed in cream or béchamel sauce when the spores were heated for 10 min at 90 °C in these two products. This was attributed to the protective effect of fat in the cream and the ingredients in the béchamel sauce. The study provides some insight into the potential microbial (stability and safety) consequences of the current trend towards milder heat treatments which is being pursued in the food industry.

Screening of different stress factors and development of growth/no growth models for Zygosaccharomyces rouxii in modified Sabouraud medium, mimicking intermediate moisture foods (IMF).

The microbial stability of intermediate moisture foods (IMF) is linked with the possible growth of osmophilic yeast and xerophilic moulds. As most of these products have a long shelf life the assessment of the microbial stability is often an important hurdle in product innovation. In this study a screening of several Zygosaccharomyces rouxii strains towards individual stress factors was performed and growth/no growth models were developed, incorporating a(w), pH, acetic acid and ethanol concentrations. These stress factors are important for sweet IMF such as chocolate fillings, ganache, marzipan, etc. A comparison was made between a logistic regression model with and without the incorporation of time as an explanatory variable. Next to the model development, a screening of the effect of chemical preservatives (sorbate and benzoate) was performed, in combination with relevant stress factors within the experimental design of the model. The results of the study showed that the influence of the investigated environmental stress factors on the growth/no growth boundary of Z. rouxii is the most significant in the first 30-40 days of incubation. Incorporating time as an explanatory variable in the model had the advantage that the growth/no growth boundary could be predicted at each time between 0 and 60 days of incubation at 22 °C. However, the growth/no growth boundary enlarged significantly leading to a less accurate prediction on the growth probability of Z. rouxii. The developed models can be a useful tool for product developers of sweet IMF. Screening with chemical preservatives revealed that benzoic acid was much less active towards Z. rouxii than sorbic acid or a mixture of both acids.

Monitoring the intracellular pH of Zygosaccharomyces bailii by green fluorescent protein.

It is generally known that intracellular pH (pH(i)) plays a vital role in cell physiology and that pH(i) homeostasis is essential for normal cellular functions. Therefore, it is desirable to know the pH(i) during cell life cycle or under various growth conditions. Different methods to measure pH(i) have been developed and among these methods, the use of pH-sensitive green fluorescent protein (GFP) as a pH(i) indicator is a promising technique. By using this approach, not only can more accurate pH(i) results be obtained but also long-term experiments on pH(i) can be performed. In this study, the wild type Zygosaccharomyces bailii, a notorious food spoilage yeast, was transformed with a plasmid encoding a pH-sensitive GFP (i.e. pHluorin), enabling the pH(i) of the yeast to be determined based on cellular fluorescent signals. After the transformation, growth and pH(i) of the yeast were investigated in four different acidic conditions at 22°C during 26days. From the experimental results, the transformation effectiveness was verified and a good correlation between yeast growth and pH(i) was noticed. Particularly, it was observed that the yeast has an ability to tolerate a significant pH(i) drop during exponential phase and a subsequent pH(i) recovery in stationary phase, which may underlie the exceptional acid resistance of the yeast. This was the first time that a GFP-based approach for pH(i) measurement was applied in Z. bailii and that the pH(i) of the yeast was monitored during such a long period (26days). It can be expected that greater understanding of the physiological properties and mechanisms behind the special acid resistance of the yeast will be obtained from further studies on this new yeast strain.

Validated empirical models describing the combined effect of water activity and pH on the heat resistance of spores of a psychrotolerant Bacillus cereus strain in broth and béchamel sauce.

The major objective of this study was to evaluate and model the combined effect of the water activity (a(w)) and pH of the heating menstrum on the heat resistance of spores of a psychrotolerant Bacillus cereus strain isolated from béchamel sauce. Two models, a quadratic polynomial equation and a reparameterized function, were assessed for their ability to describe the combined influence of a(w) and pH on the D(85°C)-values of the B. cereus isolate in tryptone soy broth. The performance of the models was validated by challenging the models with data independently collected in broth and béchamel sauce. Both models were found to adequately describe the validation data obtained in broth. However, it was determined that in béchamel sauce the predictions of the polynomial function not only showed bias (bias factor = 1.156) but were also fail-dangerous, as they deviated from the validation data by 17.2%. The reparameterized function was determined to be a good predictor of the D(85°C)-values in béchamel sauce as it showed no bias (bias factor = 1.033) and its predictions differed by only 7.9% from the validation data. The reparameterized function can be used to provide estimates of the minimum processing conditions required to achieve desired levels of spore inactivation within the a(w) and pH ranges studied and to determine the potential changes in heat resistance of B. cereus spores when a(w) and pH are changed, for example, during product reformulation. As validation of heat resistance models is rarely performed, let alone in actual food products, the models evaluated and validated in this study (in particular the reparameterized function) are of immediate relevance to the food industry.